A remaining question concerning cII is whether they selectively support collective goals. Gollwitzer and colleagues (2008) noted: “Because imple-mentation intentions are subordinate to goal intentions, they should operate in the service of meeting their superordinate goal intentions; they should not be mechanized as plans that influence behavior, regardless of the state of the superordinate goals” (p. 330). In line with this claim, past research has demonstrated that implementation intentions impact behavior when com-mitment to the superordinate goal is sufficient, but not when comcom-mitment is low (e.g., Orbell, Hodgkins, & Sheeran, 1997; Sheeran, Webb, & Gollwitzer, 2005, Study 1) or the goal has not been activated (Sheeran et al., 2005, Study 2). cII are theorized to support collective goals held by a group.
When a collective goal is cooperative and directed towards the group, it cannot be achieved when encountering a non-group member. In this case, cII should therefore not impact one’s behavior. An alternative still to be tested, however, is that cII support general cooperation, for instance by priming other-related concepts (Drouvelis, Metcalfe, & Powdthavee, 2010;
Wong & Hong, 2005). If this was the case, cII should increase cooperation, regardless of whether it serves the group’s goal or not.
To pit these explanations against each other, a task is needed where act-ing towards the collective goal can be distact-inguished from actact-ing towards the individual goal. This is the case when collective and individual goals are opposed and striving for one hampers the other. Conflicting individual and collective goals naturally arise when performingmixed motive tasks (Dawes, 1980; Hardin, 1968; van Vugt, 2009) from McGrath’s (1984) negotiate quad-rant. In mixed motive tasks, everybody is better off when everybody coop-erates, but one individual is better off when he or she defects. Examples include using natural resources such as fresh air, contributing to commons such as public broadcasting, or actively supporting institutions such as going to vote. Everybody can enjoy the benefits of clean air, public
broadcast-59
ing, and a democratically elected government, even if he or she excessively pollutes the air, does not pay fees, and does not go to vote. However, if everybody pollutes the air, refuses to pay fees, and does not go to vote, the commons cease to exist and nobody can enjoy the benefits. Despite the differences between these tasks, they all pose a conflict between individual self-interest and collective interest.
Among mixed motive tasks, social dilemma games make this conflict very explicit (Dawes, 1980; Messick & Brewer, 2005; J. M. Weber, Kopelman, &
Messick, 2004). In this type of economic game, the individual goal of making maximal profit is an obstacle to the collective goal of securing common profit: Whenever one acts in line with one’s short-term personal profit goal, one hampers the collective profit goal. Payoff-matrices show this conflict explicitly, which makes it even more difficult to resolve. Indeed, even when committed to the collective goal of securing common, long-term benefits, individuals often act selfishly and in line with their individual goal (Komorita
& Parks, 1995). Sejits and Latham (2000) even noted: “A social dilemma appears to be a boundary condition for the normally positive effect of group goal setting on group performance” (p. 104). In short, it is difficult to prioritize the collective goal over the conflicting individual goal in social dilemma situations.
This conflict between individual and collective goals is even stronger in dilemma games with monetary incentives and when participants do not in-teract repeatedly (i.e., inone-shot games). When decision-dependent mon-etary incentives are offered, the task poses a real conflict to participants (Smith, 1976) and prioritizing group welfare on the spot becomes even more difficult than in hypothetical tasks. Similarly, in one-shot games, a competi-tor cannot reciprocate cooperation. Reciprocity (including punishment) is known to increase cooperation in iterated dilemma games (Axelrod & Hamil-ton, 1981; Fehr & G¨achter, 2002). Thus, as reciprocity is impossible in one-shot games, defection becomes the dominant strategy as it yields greater payoffs for the individual regardless of the competitor’s choice (Camerer, 2003). In sum, one-shot dilemma games with decision-dependent monetary incentives instill a strong individual goal to defect, which makes it difficult to prioritize the collective goal to cooperate.
Past research on goal conflict and implementation intentions has demon-strated that prioritization II can resolve even strong goal conflicts (Kirk et al., 2011; van Koningsbruggen et al., 2011). The response in a prioritization implementation intention is to remind oneself of the focal goal, and the if-part refers to a situation when one easily acts against this goal. The if-then link created by the implementation intention swiftly activates the represen-tation of the focal goal as soon as the critical situation is encountered and prioritizes the focal goal over conflicting goals. In line with this idea, in a recent study (van Koningsbruggen et al., 2011, Study 1) a prioritization II increased the accessibility of goal-related concepts (dieting) when faced
EXPERIMENT 6: PILOT 61 with cues of a conflicting goal (tempting food), as indicated by more dieting-related word completions in a subsequent word-search task. Furnishing the collective profit goal with a prioritization cII should thus help one to achieve the goal to cooperate, even under the influence of a detrimental individual goal. However, this effect should be goal-dependent: The collective goal to cooperate refers to one’s group and thus cannot be achieved when faced with a non-group member. Thus, when facing a non-group member, cII should not increase cooperation.
To test these predictions, I developed a dilemma task based on previous research (Sheldon & Fishbach, 2011): Participants assumed the role of the CEO of an airline that was part of an alliance with other airlines (group).
The CEO’s task was to make pricing decisions (standard or discount) for a number of different routes, each also serviced by another airline. The other airline on each route was either part of the alliance or not, and both airlines’
pricing decisions impacted each other in a social dilemma fashion. Partici-pants were paid according to one of their decisions (monetary incentive) and did not learn about the other participants’ decisions until after the exper-iment (i.e., played multiple one-shot dilemma games). At the beginning of the experiment, the dilemma structure was explained to participants in de-tail to implicitly evoke both the collective goal and the conflicting individual goal. To test whether cII help prioritize the collective goal and increase co-operation, each participant then either formed a cII, a neutral control plan, or an II. Then, participants made pricing decisions for routes serviced by their airline and another alliance member followed by decisions for routes serviced by their airline and a non-alliance member. I expected cII partici-pants to make more cooperative choices than II or control participartici-pants, but only when faced with an alliance member. A pilot study was run to ensure that the individual profit goal was sufficiently strong to pose an obstacle to collective goal striving and that the group (alliance) was meaningful to participants (i.e., decreases defections). After reporting this pilot study, I will turn to the main experiment.
Experiment 6: Pilot
Method
Participants and design. Twelve students from the University of Kon-stanz (4 female, mean age = 22.83 years,SD = 2.79) participated in return for decision-contingent payment (see below) and earned 3.29 e on average (SD = .95). The pilot was run to ensure that the dilemma task provided a sufficient incentive for uncooperative choices (i.e., defection) and whether the group (alliance) was meaningful to participants. The pilot thus followed a 2 (competitor: alliance vs. non-alliance) within design.
Procedure. After giving informed consent, all participants learned that the study concerned economic decision making and that they would receive payment according to their decisions: At the end of the experiment, one trial would be chosen randomly, their decision in this trial would be matched with the decision of another, actual participant, and they would receive their decision-payoff in Euro. To ensure that participants knew that actual decisions were used, it was made clear that the results of other participants were available as printouts12 and there was no deception concerning the payoff. It was implied that participants assume different roles (represented different airlines) to provide a meaningful competition context.
Actually, all participants assumed the role of the International Airline CEO, and learned that it was important for this study to remember this airline name well and were prompted to type it (free recall). Participants then learned that they were to decide on the pricing of their tickets for differ-ent routes their airline services (see Sheldon & Fishbach, 2011, for a similar task): They can choose either standard pricing (viz., cooperate) or discount pricing (viz., defect). However, each route is also serviced by another air-line that also chooses between the two pricing options, and the outcome of both airlines’ decisions influence each other. An example payoff-matrix was provided (see Figure 8a) and explained as follows: If both choose standard-pricing, both earn well (A, a). However, if International Airline chooses discount pricing while the other airline chooses standard pricing (B, a), In-ternational Airline attracts more passengers and thus makes greater revenue.
The other airline earns less because fewer passengers use its service. The opposite situation (International Airline cooperates and the competitor de-fects; A, b) was also explained: In this case, International Airline earns less and the competitor earns more. Lastly, if both airlines choose discount pric-ing (B, b), the numbers of passengers remains stable but, due to the reduced price, both earn less than they would if both chose standard pricing. It thus is more profitable for the individual to choose the discount price (viz., de-fect) regardless of the competitors decision (B,a>A,a; B,b>A,b) but both competitors are better off when both choose standard pricing (viz., cooper-ate) instead of discount pricing (defect; B,b< A,a). To further emphasize the mixed motive structure of the task, the four outcome situations were summarized on a slide (A,a; B,a; B,b; A,b). Four questions followed that presented participants with hypothetical decision situations (e.g.,Consider the following situation: You choose discount pricing and the other airline chooses standard pricing. How much do you earn?) The situations of mu-tual cooperation, mumu-tual competition, and one competing while the other cooperates were included in the examples. Participants could only continue if they entered the correct answer and were otherwise prompted to correct
12For the first participants, actual decisions from voluntary pretesters blind to hypothe-ses were used.
EXPERIMENT 6: PILOT 63
Figure 8: Payoff matrices (Experiment 6).
their response; if needed, the experimenter reiterated the instructions. All participants can thus be expected to have implicitly formed the individual goal to defect and the collective goal to cooperate through the instructions.
Next, the group (alliance) was introduced: International Airline, Air Oceanea, and Metropolis Airways founded theFlug-Allianz in order to mar-ket residual ticmar-kets.13 Participants were informed that the alliance was im-portant to the task, asked to memorize the alliance name, and type it (free recall). This procedure has been shown to create a meaningful group (Pinter
& Greenwald, 2011). An example slide with the alliance situation was pre-sented (see Figure 8b) and explained thoroughly: The alliance-revenue for each connection is divided between the two airlines servicing the respective route 50/50. When both airlines choose standard pricing (A,a), the alliance has a lot of tickets to market and makes a lot of revenue. Discount pricing by one airline (A,b; B,a) reduces the number of residual tickets and the al-liance’s revenue. When both airlines choose discount pricing (B,b), there are no tickets left for the alliance to market leading to no alliance revenue. The pattern of the alliance payoffs thus reflected the joint payoff of both airlines.
However, the matrices were constructed so that actual payoffs were identical to the non-alliance matrices (seePayoff-matrices). Again, participants had to respond to four questions correctly before they could continue with the experiment.
Eight alliance trials followed (see Payoff-matrices). Each payoff matrix was used once with each competitor and the order was randomized. Af-ter a 30 sec break, participants worked on eight trials of the no-alliance task, allegedly against two non-members (Fly Jet and City Connext). To maintain one-shot dilemmas, participants did not learn about other partic-ipants’ decisions until after the experiment. Finally, participants provided
13To prevent any previous association, all airline names, the alliance, and the 3-letter airport codes were invented and the list of International Air Traffic Administration (IATA, 2012, retrieved from http://de.wikipedia.org) codes was checked to ensure that no respec-tive companies or airports existed.
demographic information and were debriefed. The computer program au-tomatically printed a sheet with the participant’s decisions, a randomly selected decision to be implemented (i.e., to be paid to the participant), and a randomly selected past participant whose decision on this trial served as the competitor’s choice. Decisions in this trial were checked against the payoff-matrix and the participant was compensated accordingly.
Payoff-matrices. Four payoff-matrices were constructed for the deci-sion task as follows: Cooperation-cooperation payoffs (a, A) ranged from 4 to 7; this difference was deemed sufficiently small to prevent high-stakes effects (Burton-Chellew & West, 2012). All other payoffs in the respective matrix were determined by subtracting a fixed amount (for the player: B,a +2, A,b -3, B,b -2). Thus, payment differences were held constant across payoff-matrices to keep the temptation to defect constant (Smith, 1976).
Alliance matrices were constructed by subtracting equal amounts from both competitors in the respective field of the payoff matrix. As the alliance pay-offs were divided equally between both airlines servicing the route (50/50), this left the economic payoff unchanged. Only the alliance context differed between both types of trials. Moreover, in order to prevent effects from making the same decision repeatedly, each decision slide indicated a differ-ent flight route by two fictitious 3-letter airport codes (e.g., STB-LMT).
Dependent measures. Dependent measures were the number of de-fections (i.e., number of trials where discount pricing was chosen) in both alliance and non-alliance trials.
Results
All participants responded to the manipulation check items correctly. On average, participants defected more than they cooperated (M = 11.83 out of 16 trials, SD = 3.13). This indicates that the task provided a suffi-cient incentive to act selfishly (defect). To test whether group member-ship affected participants’ pricing decisions, the defection score was entered into a repeated-measure ANOVA with competitor (alliance member vs. non-alliance member) as within-factor. As expected, participants defected less when faced with an alliance member (M = 5.25 out of 8 trials, SD = 2.30) than when faced with a non-alliance member (M = 6.58 out of 8 trials, SD = 1.38), F(1,11) = 4.63, p = .05, part. η2 = .30. In line with this, participants reported medium-to-high identification with the Flug-Allianz, M = 4.29,SD= 1.27. Together, the findings confirm that the task provides a high incentive to defect and that the alliance is a meaningful group for participants.
EXPERIMENT 6: MAIN EXPERIMENT 65
Experiment 6: Main Experiment
To test whether collective implementation intentions merely increase gen-eral cooperation or selectively support collective goal striving, Experiment 6 used this paradigm and equipped participants either with cII, II, or con-trol instructions. If cII merely promote cooperation, they should lead to more cooperative decisions no matter whether the competitor is an alliance member or not. On the other hand, if cII selectively support collective goal striving, they should only promote cooperation when competing with an alliance member, as the collective goal only applies in this context. When competing with a non-alliance member, the collective goal does not apply (i.e., cooperation does not help the group) and cII should not promote co-operation.
As explicit self-report measures of commitment are not a reliable indica-tor of the strength of implicit goals in dilemma situations (Bargh, Gollwitzer, Lee-Chai, Barndollar, & Tr¨otschel, 2001, Experiment 2), I used a different research strategy: In addition to forming implicit collective and conflicting individual goals through the task instructions, participants either formed the individual goal or the collective goal explicitly and I measured goal com-mitment to the respective goal after task performance. By manipulating whether the individual or the collective goal is made explicit, the present experiment can also show whether explicitly set collective goals might be strong enough to foster cooperation.
Method
Participants and design. One hundred and thirty-four Konstanz students (80 female) with a mean age of 20.90 years (SD = 2.58) participated in re-turn for a decision-dependent monetary incentive (as in the pilot study) and earned 4.51e on average (SD = 2.20). The experiment followed a 2 (com-petitor: alliance member vs. non-alliance member) × 3 (implementation intention: control vs. II vs. cII)×2 (explicit goal: individual vs. collective) mixed design with competitor as within-participant factor.
Procedure. The procedure of the main experiment was largely the same as in the pilot study, with the following exceptions: To manipulate the explicit goal, participants either received the explicit individual goal “I want to maximize International Airline’s revenue” after learning the non-alliance task or the explicit collective goal “We want to maximize Flug Allianz’ rev-enue” after the learning the alliance task. To make sure that all participants equally took time to think about the respective task, participants who did not receive a respective explicit goal at this point were instructed to reflect on how best to decide.
Moreover, before working on the decision trials, participants received a printed training sheet to manipulate the implementation intention factor.
Participants in the cII condition set the collective if-then plan “And when we are about to make our decision, then we will make sure that Flug Alliance receives the most;” participants in the II condition set the individual if-then plan “And when I am about to make my decision, then I will make sure that International Airline receives the most.” To minimize the differences between conditions, control participants received the neutral control plan
“When a decision screen appears, then a decision has to be made.” This plan was also phrased in the if-then format, also referred to deciding and to the task, but referred to neither the individual nor the group.
After finishing with the decision task, participants answered several ques-tionnaires concerning their commitment to the explicitly set goal (e.g., “This is a goal to shoot for”, 1not at all – 7completely, Cronbach’sα= .70, Klein, Wessen, Hollenbeck, & DeShon, 2001), commitment to their plan (e.g., “It is important to me to fulfill my plan”, 1 not at all – 5 very much, Cron-bach’s α = .86), and group identification (as in Experiment 1, Cronbach’s α = .83). Lastly, participants played a hypothetical trust game (adapted form Fischbacher, G¨achter, & Fehr, 2001). Each participant learned that International Airline (their airline) had e 10k that can be invested in a common project account with Metropolis Airways (an alliance member) or be kept in one’s own account. Metropolis Airways would also have this choice. Money in one’s own account was ones to keep and safe. All the contributions to the common project account would be added, and each air-line would receive 75 % of this sum. Investing was thus a good opportunity to make money if both contributed, but required trusting the other airline to contribute equally. Importantly, this game is structurally equivalent to the dilemma game played in the main experiment but relies more on trust.
It is thus well suited to examining whether cII led to generalized trust or created a situation-specific if-then link. Participants indicated how much of theire 10k they would transfer to the project account. Next, the same hypothetical game was played against a non-alliance member (Fly Jet).
Results and Discussion
Manipulation checks. All participants responded to the manipulation check questionnaires correctly, reported medium to high group identifica-tion,M = 4.57,SD= 1.06, commitment to their plan,M = 3.56,SD = .94,
Manipulation checks. All participants responded to the manipulation check questionnaires correctly, reported medium to high group identifica-tion,M = 4.57,SD= 1.06, commitment to their plan,M = 3.56,SD = .94,